McGarry (1983)
McGarry (1983)
313
Liquid vapor pressures of 72 substances are available over pressure ranges extending from about 1 kPa up to
the particular critical pressure. For correlation of these data the Wagner equation yields an average root mean
square percentage error of 0.09 % . The results obtained allow the formulation of constraints applying to the behavior
of vapor pressure for values of the reduced temperature in the range 0.5 to 1.O. Using these constraints Wagner
coefficients were generated and are presented for a further 179 pure substances for which only limited range data
are available (the limits are usually about TR = 0.55 and 0.65). It is considered that these coefficients will yield
vapor pressures in the range TR = 0.5 to 1.0 which are suitable for process design work.
Introduction
The calculation of accurate liquid vapor pressures is vital
for the success of many chemical engineering design
techniques. Numerous equations have been proposed for
thia purpose, and eight of these together with modifications
of four of them are considered in this paper.
The widely used Frost-Kalkwarf equation (Frost and
Kalkwarf, 1953) is not included in these eight because it
is not explicit in vapor pressure. It may be worth mentioning, however, that this equation was fitted to the 14
sets of full range data initially involved with an average
% RMS error of 0.12.
The selected eight equations may be divided into two
groups on the basis of the method by which the constants
(coefficients) of the equations are calculated. (Group I):
Constanta are calculated by the regression of experimental
vapor pressure. The Antoine, Wagner, and Thomas
equations belong to this group (Antoine, 1888 Wagner,
1973: Thomas, 1976). Representation by Chebyshev
polynomials also falls into this group (Ambrw et al., 1970).
(Group 11): Constants are calculated from critical temperature and pressure together with one or two further
basic data. The Riedel, Miller, Thek-Stiel, and GomezThodos equations belong to this group (Riedel, 1954:
Miller, 1965: Thek and Stiel, 1966: Gomez-Nieto and
Thodos,1977). It seems obvious that it could be beneficial
to use regression of data for evaluation of the constants
of the equations of the second group.
The Equations Involved
Antoine Equation.
In P = A - T+C
This equation can yield highly accurate values of vapor
pressure in a pressure range of about 1to 200 P a . Values
of constants valid for this range and yielding vapor pressure in millimeters of mercury are available for many
substances (Reid et al., 1977; Boublik et al., 1973). The
equation was developed from the Clausius-Clapepon
equation and is used extensively.
Wagner Equation.
1
In PR = -[A(1 - TR) + B(l - TR)'" C(l - T R ) ~
TR
D(1 - T R ) ~(2)
]
Statistical considerationswere involved in the development
of this equation, which was initially derived to describe
the vapor pressures of argon and nitrogen from the triple
where
(6)
(7)
019~-4305ia3ii
122-0313$01.50io @ 1983 American Chemlcal Society
314
No. 2, 1983
B = -36Q
C = 42Q + a,
D = -Q
Q = 0.0838(3.758 - a,)
(12)
where a, is called the Riedel parameter and is defined by
a, = d (In PR)/d (In TR) at TR = 1
(13)
Antoine
Wagner
Thomas
Chebyshev
Riedel
Miller
Thek-Stiel
Gomez-Thodos
eq no.
1
2
3
5
TC,pC,
20
25
where
A = 0.4835 + 0.4605h
TBR In (1O1.325/Pc)
h=
1 - TBR
(17)
(18)
Thek-Stiel Equation.
where
f(TR) =
1
1.14893 - - - 0.11719TR-0.03174T~~
- 0.375 In TR
TR
(22)
(23)
(24)
a, is again obtained from a knowledge of normal boiling
temperature, and h is given by eq 18.
Gomez-Thodos Equation.
B = 1 . 0 4 2 ~- ~0.462844
~
C = 5.2691 + 2.0753A - 3.1738h
(25)
where
7.0109
380900
C = 2.4186 - h
he(123.21/h)
+
DE-
a,
A = -(B
(26)
+ BC
7
+ D)
(29)
TB
Tc, P c , TB,AHB
Tc, P c , TB,LY,
where
fitted constants A , B, C
Tc, Pc, fitted constants A , B, C, D
fitted constants A , B, C
TL,TH,fitted constants a, to as
Tc,P c , TB
T,,
T,,
T,,
T,,
T,,
fitted constants Po A , B,
P,, fitted constants A , B,
fitted constants Po A, B
fitted constants Pc, A , B,
P,, fitted constants A , B,
C, D
C, D
C, D
C, D
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 2, 1983 315
Table 111. Sources of High Accuracy Data in Range
5 kPa to PC
ID
source
benzene
toluene
ethylbenzene
o-xylene
3
4
m-xylene
p-xylene
diethyl ether
acetone
cyclohexane
5
6
7
8
9
substance
water
methanol
10
11
ethanol
12
nitrogen
argon
13
14
no. of
data points
total
data points
19
19
9
20
4
9
47
20
9
17
9
12
12
20
12
25
23
33
33
29
29
29
40
47
38
49
44
48
49
49
+ (0.3)3C+(0.3)60]
(30)
Unconstrained Fit to Limited Ranges of Data
There are many substances for which data only exist
over a small range of pressure, the usual upper limit lying
in the range 100 to 200 kPa. The use of such data for
successful prediction of vapor pressures outside the particular range is obviously desirable.
Consequently, data were extracted from the lower ends
of the full ranges of data for the substances of Table 111,
and equation constants were obtained by regression on
these samples of data. Since no data near the critical
points are involved, it w i l l be evident that critical pressures
should not be subjects of iteration in eq 2, 16 and 20. By
use of these sample constants, vapor pressures were calculated for the full range of data and the root mean square
percentage deviations were obtained. The results are
presented in Table VI (Supplementary Material).
Outstandingly poor results are obtained for argon, and
this is due to the fact that the temperature range yielding
vapor pressures up to about 100 P a is too small a fraction
of the range up to the critical pressure (0.06 compared with
about one quarter for the other substances). Excluding
argon, the average root-mean-square percentage deviations
resulting from eq 1, 2 , 3 , 11, 16, 20, and 25 were 2.3, 0.3,
71.0, 9.1, 0.75, 0.20, and 8.0 respectively.
When generating Thek-Stiel eq 20 constants from the
samples of data, it was necessary to stop the iteration
procedure prematurely to avoid excessive computation.
This had the effect of preventing the equation fitting the
sample data more accurately but gave a better fit over the
full range. In consequence, the percentage deviations given
316
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 2, 1983
formula
Ar
name
argon
-5.90501
1.12627
-0.767869
co
carbon
-6.20798
1.27885
-1.34533
monoxide
carbon dioxide -6.95626
1.19695
-3.12614
CO,
cos
carbonyl sulfide -6.40952
1.21015
-1.54976
fluorine
-6.18224
1.18062
-1.16555
F,
He
helium
-3.97466
1.00074
1.50056
hydrogen
-5.57929
2.60012
-0.855060
HZ
water
-7.76451
1.45838
-2.77580
HZ 0
Ne
neon
-6.07686
1.59402
-1.06092
nitrogen
-6.09676
1.13670
-1.04072
NZ
oxygen
-6.28275
1.73619
-1.81349
0 2
CCl,F,
dichlorodi-7.01657
1.73224
-2.97909
fluoromethane
CC1,
carbon
-7.07139
1.71497
-2.89930
tetrachloride
CHClF, chlorodi-6.99913
1.23014
-2.49377
fluoromethane
CHC1,
chloroform
-6.95546
1.16625
-2.13970
CHF,
trifluoro-7.41994
1.65884
-3.14962
methane
difluoromethane -7.44206
1.51914
-2.75319
CHZFZ
CH3C1 methyl chloride -6.86672
1.5227 3
-1.92919
methyl fluoride -6.78099
CH-F
0.828379
-1.41137
CH;
methane
-6.00435
1.18850
-0.834082
CH,O
methanol
-8.547 96 0.769817
-3.10850
CH,N
methylamine
-7.527 7 2 1.81615
-4.20677
trichlorotri-7.36666
1.81971
-3.94233
C2C13F3
fluoroethane
ethylene
-6.32055
1.16819
-1.55935
1,2-dichloro-7.36864
1.767 27
-3.34295
ethane
ethane
-6.34307
1.01630
-1.19116
'ZH6
ethanol
-8.51838
0.341626
-5.73683
'ZH6'
ethyiamine
-7.20059
C,H,N
1.20679
-3.71972
dimethy lamine -7.9 02 9 5 2.81577
-6.31338
propylene
-6.64231
-1.81005
1.21857
acetone
-7.45514
1.20200
-2.43926
prop an e
-6.67833
1.15437
-1.64984
1-propanol
-8.05594
4.25183E-2 -7.51296
2-propanol
-8.1 6927 -9.4321 3E-2 -8.10040
propylamine
-7.23587
1.22853
-3.75004
trimethylamine -6.88066
1.15962
-2.18332
butene
-6.88204
1.27051
-2.26284
isobutene
-6.95542
1.35673
-2.45222
methyl ethyl
-7.71476
1.71061
-3.68770
ketone
butane
-6.88709
1.151 57
-1.99873
isobu tane
-6.95579
1.50090
-2.52717
diethyl ether
-7.29916
1.24828
-2.91931
1-butanol
-8.007 56 0.537 826
-9.34240
2-butanol
-7.80578
0.324557
-9.41265
diethylamine
-7.267 96 1.15810
-3.91125
1-pentene
-7.04875
1.17813
-2.45105
pentane
-7.28936
1.53679
-3.08367
isopentane
-7.12727
1.38996
-2.54302
2,2-dimethyl-6.89153
1.25019
-2.28233
propane
1-pentanol
-8.97725
2.99791
-12.9596
chloropenta-8.0217 2 1.54665
-3.78361
fluorobenzene
pentafluoro-7.79730
1.35271
-3.50409
benzene
1,2,4,5-tetra-7.79740
1.57406
-3.82060
fluorobenzene
benzene
-6.94739
1.25253
'6
H6
-2.53686
cyclohexane
-6.96009
1.31328
-2.75683
hexane
-7.51650
1.54797
-3.38541
isohexane
-7.28750
1.29015
C6H14
-2.97853
3-methyl-7.27084
1.26113
6'
-2.81741
pentane
2,2-dimethyl-7.25933
1.69602
-3.18124
butane
2,3-dimethyl-7.27870
1.56349
-3.05387
butane
C~HIN
D
-1.62721
-2.56842
2.99448
-2.10074
-1.50167
-0.430197
1.70503
-1.23303
4.06656
-1.93306
-2.536453-2
-0.377232
pc
4857.99
3501.15
7374.99
6346.45
5214.72
230.029
1309.60
22122.3
2724.55
3399.61
5089.87
4132.03
Tc
rms %
error
150.651 0.023
132.91 0.28
approx
lowest data
69
26
84
71
304.15
378.8
144.31
5.20
33.19
647.35
44.38
126.200
154.7
384.95
0.011 530
0.18
2
0.012
4
0.26
5
7
0.27
0.026
0.7
0.57
43
0.025 1 3
0.28
0.2
0.038
0.2
217
162
64
2
14
275
25
63
54
155
-2.49466
4550.78
556.40
0.027
250
-2.21052
4983.31
369.30
0.042
170
-3.44421
-0.849379
5365.76
4840.92
536.40
299.06
0.14
0.075
0.1 215
0.2 125
-0.979495
-2.61459
-2.41700
-1.22833
1.54481
-1.22275
0.625601
5826.99
6697.18
5557.36
4596.42
8 0 8 5.0 5
7433.32
3425.71
351.54
416.27
315.0
190.53
512.64
430.0
487.7
0.066
0.11
0.079
0.019
0.12
0.073
0.32
-1.83552
-1.43530
5050.88
5362.00
282.55
566.00
0.034
0.071
0.1 1 0 5
1
260
-2.03539
8.32581
-4.33511
-0.224073
-2.48212
-3.35590
-2.70017
6.89004
7.85000
-4.33990
-2.94707
-2.61632
-1.46110
-0.751692
4869.71
6130.87
5641.37
5308.28
4605.23
4699.93
4255.76
5151.11
4742.44
4806.74
4083.96
4017.60
4007.06
4221.77
305.42
513.92
456.35
437.70
364.85
508.10
369.82
536.78
508.30
497.0
433.30
419.57
417.90
536.78
0.13
0.077
0.089
0.043
0.018
0.086
0.043
0.19
0.22
0.086
0.073
0.094
0.086
0.11
2
6
1
1
0.1
4
0.2
0.2
0.1
1
1
0.2
0.2
1
133
293
215
240
140
259
145
260
250
235
200
170
170
255
-3.13003
-1.49776
-3.36740
6.68692
2.64643
-1.17981
-2.21727
-1.02456
-2.45657
-4.74891
3790.62
3658.01
3646.10
4412.63
4189.75
3705.40
3536.85
3378.62
3385.90
3197.88
425.18
408.14
466.74
563.05
536.01
496.45
464.78
469.74
460.43
433.77
0.11
0.092
0.072
0.14
0.10
0.043
0.095
0.074
0.039
0.036
0.1
0.2
7
2
0.2
1
0.1
0.1
2
41
170
165
250
275
265
240
190
195
220
260
8.84205
-2.99849
3909.45
3235.98
588.15
570.81
0.096
0.024
-3.76856
3535.63
530.97
0.012
27
322
-2.45398
3801.01
543.35
0.017
294
-3.49284
-2.45491
-2.36767
-2.17234
-2.17642
4895.60
4075.26
3036.17
3032.52
3121.71
562.10
553.640
507.90
498.10
504.40
0.084
0.051
0.028
0.046
0.037
-0.805183
31 12.72
489.40
0.065
225
-1.57752
3145.80
500.30
0.066
235
0.6
0.8
0.7
12
10
1
2
155
175
135
91
288
200
238
0.2 290
4
309
8
288
10
293
0.2 220
1
240
1
235
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 2, 1983 317
Table VI1 (Continued)
approx
rms % lowest data
formula
C,H,F,
C,H,
C,H,,
C,H,,
C,H,,
C,H,,
C,H,,
C,H,,
C,H,,
C,H,,
C,H,,
C,H,,O
3123.89
566.52
0.019
313
-2.79168
-3.20243
-2.42625
-1.10598
4106.45
2732.37
2891.48
2949.13
591.72
540.10
540.64
531.17
0.038
0.063
0.022
0.057
6
0.1
1
1
309
240
265
250
-2.23048
-2.85992
-2.40376
-2.78710
-4.44565
-2.22001
3601.90
3732.98
3536.78
3513.00
2482.82
2561.55
617.12
630.25
616.97
616.15
568.81
543.96
0.041
0.026
0.014
0.021
0.10
0.11
6
6
6
6
0.1
1
330
337
332
331
260
265
-3.59254
2873.76
652.5
0.13
0.1 325
pentafluorotoluene
toluene
heptane
3-ethylpentane
2,2,3-trimethylbutane
ethylbenzene
o-xylene
m-xylene
p-xylene
octane
2,2,4-trimethylpentane
1-octanol
-8.04616
1.43971
-3.76736
-3.00179
-7.28607
-7.67468
-7.58305
-7.22017
1.38091
1.37068
1.58587
1.44914
-2.83433
-3.53620
-3.567 3 2
-3.11808
-7.48645
-7.53357
-7.592 22
-7.63495
-7.87867
-7.38890
1.45488
1.40968
1.39441
1.50724
1.32514
1.25294
-3.37538
-3.10985
-3.22746
-3.19678
-3.78494
-3.16606
-9.71763
4.22514
Waring (1954) observed that for a wide variety of substance a plot of the left-hand side of eq 32 against reduced
temperature exhibits a minimum value at a value of TR
in the range 0.80 to 0.85. Ambrose et al. (1978) observed
that the minimum occurs at a somewhat lower value of TR
for very low-boiling substances and at about TR = 0.95 for
alcohols.
Differentiation of the left-hand side of eq 32 and
equating the result to zero will yield the minimizing value
of TR. If the Wagner equation is used, the following
equation is obtained for the minimizing value of TR
0.75B(1 - T R ) ~ 'i
. '6C(1 - TR) 300(1 - T R )=~0 (33)
error
-12.9222
Tc
name
PC
frattional
temp, temp
value
K - range
max % dev
rms % dev
substance ID
sample
all
1
2
3
4
5
6
7
8
9
10
11
12
13
14*
14
av value
(excluding 14*)
0.002
0.019
0.016
0.014
0.010
0.006
0.028
0.017
0.015
0.014
0.018
0.015
0.043
0.023
0.028
0.018
0.29
0.13
0.05
0.11
0.12
0.07
0.23
0.11
0.12
0.19
0.50
0.20
0.10
0.99
0.21
0.17
-0.55
-0.34
-0.22
-0.24
0.30
-0.17
0.33
0.43
0.33
-0.33
-1.1
-0.54
-0.12
-1.6
-0.42
500
480
600
500
560
500
400
480
470
490
480
490
120
130
140
0.24
0.27
0.28
0.28
0.28
0.29
0.27
0.28
0.24
0.26
0.22
0.27
0.23
0.06
0.24
that the values fell into quite narrow ranges. With the
exception of alcohols, it was possible to classify the ranges
of TR according to the normal boiling temperatures (TB)
of the substances. In fact, for 50 K C TB C 100 K the range
is 0.70 to 0.77; 100 K C TB C 273 K, 0.78 to 0.85; 273 K
C TB, 0.82 to 0.88; and for alcohols, 0.90 to 0.98. Substances with values of TB below 50 K do not exhibit such
minima.
When fitting the Wagner equation to limited range data,
the values of the coefficients were restricted to those sets
yielding a solution for eq 33 that lies within the appropriate
range for the substance.
Second Constraint. At low pressures the term LvI/(Zv
- 2,) in eq 31 and 32 varies only weakly with temperature.
Assuming this function is constant and integrating eq 31
yields
In P34= A - (B/T)
(34)
The constants A and B may be obtained from vapor
pressure data at T = TCand T = 0.7Tc. Ambrose et al.
(1978) show that In
evaluated at TR = 0.95 falls
into a narrow range for many substances. Making use of
the Pitzer acentric factor (w), eq 34 becomes
In ( P 3 4 / P ~
= )- 0.28278 (1 w ) for TR = 0.95
(35)
318
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 2, 1983
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 2, 1983 319
320
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 2, 1983
a)
E
0
aJa
a ) a ) a ) a ) a ) a ) a ) w a ) a ) w
""C
cow
C C C E C E C C C C C
E
0 c
0 E
0 E
0 c
0 E
0
E C
L ~ S G ~ S NrrWE0 c0 c0 c0 c0
a)aa
E C C W
0 0
E
~ r i l " c j r l0
crr E
0 E
0 0
cei
a ) a ) a ) a )
N M E C E C
rr-l"cjc
0 E
0 c
0 0
C
2m
N C
0~
<
mdwON0w00mwm0wwmwwwmwww~w0rimwwwwwwwwwwoiwwwwwwwwww
rirrNC\l
rim
0.1
Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 2, 1983
wwwwwaww~wwwwwwwwwwmwwoF
@I
FJ
5-
+.a
.-0
Y
w w a w
9999
42
.
I
2
2
2
'i
.Y
.y,
.w
42
I
U
(37)
where g is a constant.
At low values of TR vapor pressures are low and eq 32
approximates to
d(ln P )
RP-=AH
dT
Combining eq 37 and 38
RP
d(ln PR)
(39)
g = 1 - TR)0.375 dT
In terms of Wagner coefficients this becomes
-R Tc
[A BV"(1.5 - 0.5Y) + CV(3 g=
(1 - T R ) ~ ' ~ ~ ~
2Y) + D P ( 6 - 5Y)] (40)
where Y = 1 - TR.
Ambrose et al. (1978) suggest that the coefficients of a
vapor pressure equation should be restricted to those sets
which yield little differences in g at TR = 0.5 and 0.6. The
constants of the fitted Wagner equation were used to
calculate the value of g at these values of TRfor the 72
substances of Table VII.
The ratios g(TR= 0.5) to g(TR = 0.6) fell into the following narrow ranges which constitute the third constraint
employed for fitting the Wagner equation to limited range
data. For TB C 100 K, the range is 1.06 to 1.20; 100 K C
TB C 273 K, 1.01 to 1.06; 273 K C TB, 0.99 to 1.03; for
alcohols, 0.98 to 1.06. There were no exceptions to these
ranges.
By use of the same samples of data that were involved
in Table VI and the three constraints detailed above,
values of the Wagner equation constants were obtained
by a computer program utilizing NAG routine E04VBF
for constrained minimization of an objective function.
These constants were then used to calculate vapor pressures up to the critical pressure, and the root mean square
percentage deviations from the full range data were obtained. The results are given in Table VIII.
The last entry in the table relates to an argon data
sample which represenb a fraction of the full-range data
having a similar size to the other entries. While the constraints caused a dramatic reduction of the overall deviation obtained from the smaller sample (0.99 cf. 81) the
larger sample yielded an overall deviation (0.21) more in
keeping with those of the other substances.
It will be seen from Tables VI (Supplementary Material)
and VI11 that the use of constraints improves full range
prediction using only limited range data. The uncon-
c,
321
322 Ind. Eng. Chem. Process Des. Dev., Vol. 22, No. 2, 1983